InstructPLM-mu: protein fitness prediction from structure instructions
Junde Xu, Yapin Shi, Lijun Lang, Taoyong Cui, Zhiming Zhang, Guangyong Chen, Jiezhong Qiu, Pheng-Ann Heng · 2026
Read InstructPLM-mu: protein fitness prediction from structure instructions by Junde Xu, Yapin Shi, Lijun Lang, Taoyong Cui, Zhiming Zhang, Guangyong Chen, Jiezhong Qiu, Pheng-Ann Heng (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13652651.txt.
Description
Junde Xu and co-authors ask whether a sequence-only protein model can learn structure on the job, in a 2026 paper. Models that read both sequence and shape predict mutation effects well but cost a fortune to train from scratch. Their framework, InstructPLM-mu, fine-tunes the sequence model ESM2 while feeding it structure as instructions.
The tuned model matched the far pricier multimodal ESM3. Three ways of fusing the shape data, and the tuning recipes around them, moved accuracy a lot, so the details matter. The paper is a practical guide for labs that want structure-aware predictions without a supercomputer budget.
Catalog record
| CREATOR | Junde Xu, Yapin Shi, Lijun Lang, Taoyong Cui, Zhiming Zhang, Guangyong Chen, Jiezhong Qiu, Pheng-Ann Heng |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |